CHANGELOG.md

1.1.0 (Unreleased)

NEW FEATURES:

• API Framework: Replicator now runs as an RPC server and implements an HTTP API framework. [GH-186]
• Scaling Provider: Replicator now implements a cloud provider model to support cluster scaling operations across multiple cloud providers. [GH-203]
• OpsGenie Notifier: Replicator now supports OpsGenie as a backend notifier. Thank you to @vladshub. [GH-225]
• Worker Pool Configuration In Consul: Replicator now supports the ability to store full worker pool configuration in the Consul Key/Value store allowing for minimal configuration to be stored in the node meta configuration parameters. [GH-204]
• replicator_retry_threshold: A new config flag added to allow configuration for retrying job scaling activites before entering failsafe. Thank you to @djenriquez [GH-261]

IMPROVEMENTS:

• Replicator configuration now takes into account Nomad ACL token and Consul HTTP scheme. GH-273

BUG FIXES:

• Filter out Nomad jobs that do not include an update stanza to prevent segmentation violations when attempting to verify job scaling activities. Thank you to @burdandrei. [GH-209]
• Update init command to generate correct cluster_scaling meta tags. Thank you to @burdandrei. [GH-210]
• Pass job scaling verification and processing methods the job ID rather than name to handle jobs where the name differs from the ID. Thank you to @burdandrei. [GH-212]
• Prevent processing of Nomad system jobs. [GH-220]
• Prevent panic when Replicator fails to determine the IP address of a node during worker pool scale-in operations. Thank you to @burdandrei. [GH-222]
• Prevent errors when Replicator attempts a cluster scale-in operation while the prioritized scaling metric is ScalingMetricDsik. [GH-240]
• Increase timeout for ASG scaling result. [GH-258]
• Use time.NewTicker rather than time.Tick for garbage collection issues and timeouts. [GH-262]
• Continue on with node termination even if the ASG detach fails. [GH-269]
• Fix issue caused by changes Nomad 0.8 where deployments can now return null. [GH-270]

IMPROVEMENTS:

• Remove explicit min/max thresholds for cluster scaling in favor of dynamic threshold discovery in the cloud provider. [GH-211]
• Remove all internal constants used in favor of exported constants from the Nomad structs package. [GH-213]
• Job scaling API calls now allow stale reads to reduce load on the Nomad leader. [GH-224]
• Replicator now implements concurrent scaling limits on both job and cluster scaling operations. [GH-223, GH-221]
• Replicator now implements a status/leader API endpoint for retrieving Replicator leader information. [GH-242]
• Logging messages to include date, time and zone [GH-254]

1.0.3 (22 September 2017)

BUG FIXES:

• Reset accumulated job scaling values during each scaling evaluation. Thank you to @comebackoneyear and @lifesum. [GH-202]
• Average job utilization from all allocations, not just the last one processed. Thank you to @comebackoneyear and @lifesum. [GH-202]

1.0.2 (15 September 2017)

BUG FIXES:

• Improve and fix node protection mechanism. [GH-197]

1.0.1 (14 September 2017)

BUG FIXES:

• Reset failure count when exiting failsafe [GH-190]
• Retry when Nomad returns an empty deployment ID [GH-191]

IMPROVEMENTS:

• Improve accuracy and detail of cluster_scaling telemetry metrics. [GH-194]
• Improve accuracy and detail of job_scaling telemetry metrics. [GH-193]
• Improve logging output of cluster scaling operations when logging set to INFO. [GH-195]

1.0.0 (11 September 2017)

IMPROVEMENTS:

• Nomad JobScaling now uses meta parameters within the job file for configuration [GH-139]
• Replicator now tracks scaling deployment to confirm success [GH-146]
• Introduce Node Discovery With Dynamic Refresh [GH-148]
• Replicator now supports setting failsafe per job group [GH-152]
• Replicator now supports concurrent cluster scaling of multiple worker pools [GH-157]

0.0.2 (03 August 2017)

IMPROVEMENTS:

• Replicator now supports persistent storage of state tracking information in Consul Key/Value stores. This allows the daemon to restart and gracefully resume where it left off and supports graceful leadership changes. [GH-92]
• Scaling cooldown threshold is calculated and tracked internally rather than externally referencing the worker pool autoscaling group. [GH-91]
• Failed worker nodes are detached after maximum retry interval is reached for troubleshooting. [GH-76]
• Add support for event notifications via PagerDuty [GH-66]
• Replicator will send a notification if failsafe mode is enabled [GH-131]
• New nodes are verified to have successfully joined the worker pool after a cluster scaling operation is initiated. Replicator will retry until a healthy node is launched or the maximum retry threshold is reached. [GH-62]
• Allow Dry Run Operations Against Job Scaling Documents. [GH-53]
• Add the scaling_interval config parameter to replicator agent. [GH-49]
• Increase Default Cluster Scaling Cooldown Period to 600s. [GH-56]
• Addition of CLI flags for all configuration parameters. [GH-61]
• Add support for dev CLI flag [GH-63]
• AWS TerminateInstance function now verifies instance has successfully transitioned to a terminated state. [GH-80]
• Make use of telemetry configuration by sending key metrics. [GH-85]
• Replicator now runs leadership locking using Consul sessions + KV [GH-101]
• Introduce distributed failsafe mode and new failsafe CLI command [GH-105]
• cluster-scaling-theshold parameter is now used to determine scaling safety [GH-115]

BUG FIXES:

• The global job scaling enabled flag is now evaluated before allowing job scaling operations. [GH-81]
• Prevent Divide By Zero Panic When Replicator Detects No Healthy Worker Nodes. [GH-55]
• Prevent Cluster Scaling Operations That Would Violate ASG Constraints. [GH-142]

0.0.1 (02 May 2017)

• Initial release.